anyone seen this? - idea parser getting stuck

Recently, I quite frequently see IDEA come into a mode where it flags large amounts of my source files as "unused" (yellow color), which they are definitely not. Just now it happened that IDEA even flagged an error which was not actually there.
If this happens, the only remedy is to go to the import section (thats where the unused signs usually start, but maybe its just because its the beginning of the file), and edit a line (e.g., delete and re-insert a character) such that the file gets reparsed - and voila, all is well. Almost feels like someone went to far in trying to optimize the parsing speed (although I still sometimes have to wait several seconds until the IDE becomes responsive)

this is Sun 1.4.1_01 under Linux 2.4

2 comments

Christian Sell wrote:

Recently, I quite frequently see IDEA come into a mode where it flags
large amounts of my source files as "unused" (yellow color), which
they are definitely not. Just now it happened that IDEA even flagged
an error which was not actually there.

If this happens, the only remedy is to go to the import section
(thats where the unused signs usually start, but maybe its just
because its the beginning of the file), and edit a line (e.g., delete
and re-insert a character) such that the file gets reparsed - and
voila, all is well.


I usually run IDEA under Windows 2000 (work computer) or XP (laptop),
but recently I've been running it on my own Linux computer and I've been
experiencing exactly the same problem you've described here (also Linux
kernel 2.4, Sun JDK 1.4.1_01). I've been able to get rid of the problem
by editing the file anywhere, though.

I think it smells like some kind of race condition. I may be mistaken,
but I have a feeling it seems to happen if I edit something exactly when
the reparse delay times out, so that I change some internal structure
when IDEA is trying to inspect that structure and update its
highlighting. Maybe the thread model under Linux is sufficiently
different from the Windows thread model that this race condition can
only occur under Linux.

0

I've seen this too.

I've never been able to remember a sequence of events to follow to reproduce it though.

I don't experience any apparent parsing delays however.

(Mandrake Linux 9.0)

0

Please sign in to leave a comment.